How Your Immune System Works & How to Improve It | Dr. Max Krummel

2026-08-03 - 148 min - source - Read full transcript

Andrew Huberman (host)Dr. Max Krummel

Key insights

The immune system's core job is a self-versus-non-self discrimination problem, illustrated by a WWII submarine sonar analogy.

Krummel opens with the story of submarines carrying two 'books' of engine sound profiles, one for friendly subs (don't fire) and one for enemy subs (fire). The immune system runs the same logic on molecular patterns: cells whose protein signatures match the 'self' book are left alone, and those matching a 'foreign' pattern are attacked.

self-vs-nonself

Aging degrades self-recognition because the body becomes a genetic mosaic, not because immunity simply weakens.

Skin cells alone accumulate roughly 10,000-30,000 mutations per cell per day from UV exposure. Over decades, no two cells in the body carry identical DNA anymore, so the immune system's 'self' reference book fills with noise. This makes it progressively harder to flag truly dangerous cells like early cancers, which is one reason cancer incidence rises with age.

immune-aging

The thymus is named for T-cells and its age-related shrinkage ('involution') is a second, distinct driver of immune aging.

The organ was discovered to be essential when Jacques Miller removed it from newborn mice and found they became fatally susceptible to bacterial infection and developed tumors. The thymus produces and 'educates' new T-cells early in life, then shrinks dramatically with age, cutting off the supply of fresh, correctly-tuned immune cells, which is why thymic revitalization is now a research target for both aging and cancer therapy.

immune-aging

Sleep deprivation weakens immunity through measurable, if only partly understood, physical mechanisms.

During sleep, immune cells retreat into the bone marrow and neutrophils repopulate tissues to deposit collagen as part of a nightly reparative process. Puffy under-eye bags and glassy eyes after poor sleep are literal lymphatic fluid pooling from impaired clearance. Krummel is candid that the field has solid data on these individual pieces but no unified mechanistic account of why one bad night makes a person catch a cold from someone coughing across the room.

immune-aging

The brain's insular cortex can store and later reactivate immune states as part of emotional memory.

An Israeli lab (Rossi/Roey lab) induced inflammatory bowel disease in mice, tagged the insular cortex neurons active during the flare, and later artificially reactivated just those neurons using DREADDs. The mice showed immune cell accumulation in the gut resembling an active flare, even with no new physical trigger, suggesting that recalling an emotionally charged memory can partially recall the bodily immune state that accompanied it.

mind-body-immune-connection

Cancer immunotherapy reframed immunology from a binary on/off system to a tunable, contextual one.

Krummel describes immunity pre-1990s as a 'fuel gauge': low against self, high against viruses. Discovering that blocking specific T-cell checkpoint molecules could raise reactivity just enough to let T-cells attack tumors (his own thesis experiment, which led to his advisor's Nobel Prize) revealed a wide middle zone where the immune system can quarantine, tolerate, or attack with much finer gradation than previously understood.

cancer-immunotherapy

Biological systems resist single-lever fixes, so future disease treatment likely requires sequenced, multi-step interventions rather than one-and-done drugs.

Because the body evolved to be resilient to disruption, hitting one molecular target rarely produces a clean cure and can be exploited by the disease. Krummel argues cancer treatment will increasingly require nudging tumors through a sequence of biological 'states' (e.g., stripping defenses before triggering immune attack) rather than a single drug, an approach that requires mapping cell-cell signaling networks (spatial biology) to know what order of nudges to apply.

cancer-immunotherapy

AI and machine learning are powerful for analyzing known biology but cannot generate genuinely novel discovery on their own.

Krummel distinguishes machine learning (finding statistical relationships within existing data, useful for proposing which of many possible experiments to run) from the popular expectation that AI can leap beyond the existing 'corpus' of knowledge. Real discovery, he argues, still requires physically running experiments because the answer often isn't present anywhere in the training data.

cancer-immunotherapy

Most vaccine-hesitant parents are not anti-vaccine; they are asking under-studied questions about timing and combination schedules.

Krummel argues that current vaccine schedules were built largely on convenience and single-study protocols rather than rigorous testing of alternative spacing, and that this is a legitimate, understudied question that pharmaceutical companies have little financial incentive to fund. He states he vaccinated his own children but delayed a dose by roughly a month based on his read of the underlying mouse immunology data, and calls for a dedicated public study of alternative schedules rather than dismissing the question.

vaccine-nuance-and-trust

Maternal immune activation during pregnancy is an established tool for studying neurodevelopment in animal models, which lends biological plausibility (not proof) to inflammation affecting brain development.

Researchers induce autism-like traits in mouse pups by injecting a bacterial infection into pregnant mothers, showing that maternal immune activation measurably alters offspring neurons. Krummel is explicit that this does not establish vaccines cause autism, but it does establish that immune insults during development are not biologically implausible as a factor worth researching directly rather than treating as fringe.

mind-body-immune-connection

Autoimmune conditions often reflect an evolutionarily useful immune program misapplied to the wrong context, not simple malfunction.

Krummel uses psoriasis (linked to the interleukin-17 pathway) as an example: suppressing IL-17 clears the skin but may reduce resistance to certain infections, mirroring how the sickle-cell trait persists in some populations because it confers malaria resistance despite its costs. He frames autoimmune disease sub-types as different immune 'archetypes,' which is why the same drug class (e.g., TNF blockers for inflammatory bowel disease) works well for some patients and not others.

vaccine-nuance-and-trust

Books referenced

Protocols: An Operating Manual for the Human Body - Andrew Huberman - Huberman's forthcoming first book, available for presale at protocolsbook.com; covers protocols for sleep, exercise, stress, focus, and motivation.
The Prince of Medicine - Susan P. Mattern - Huberman recommends it as background on Galen and how barbaric early medicine (surgery without anesthesia) actually was.

Media referenced

The Immune Beyond (Max Krummel's Substack) - other - Krummel's zero-cost newsletter where he writes about immunology and science communication; Huberman repeatedly recommends it.
Huberman Lab Neural Network Newsletter - other - Zero-cost monthly newsletter with podcast summaries and one-to-three-page protocol PDFs, mentioned in the closing.
Huberman Lab episode with Alex Marson - podcast - Referenced as the place to learn about engineered CAR-T cell approaches, contrasted with Krummel's discussion of natural thymic T-cell production.
Huberman Lab episode(s) with Jay Bhattacharya - podcast - Referenced when Huberman raises the question of whether vaccine spacing and combination schedules deserve more study.

Companies

Juve - Sponsor; medical-grade red light therapy devices for cellular and mitochondrial health.
8Sleep - Sponsor; smart mattress cover with temperature regulation for sleep.
AG1 - Sponsor; discusses the new AG1 Pro formulation with added creatine, calcium HMB, and zinc carnosine.
Function - Sponsor; comprehensive lab testing (160+ biomarkers) with physician-reviewed recommendations.
California Institute for Regenerative Medicine - Referenced as the state stem-cell research initiative that funded two large bond measures; Krummel notes it produced fewer clinical stem-cell therapies than originally hoped, illustrating how research timelines outpace expectations.

Techniques and frameworks

Self versus non-self discrimination (submarine sonar analogy) - Krummel's framing for how the immune system distinguishes 'self' from 'foreign' using a WWII submarine engine-sound-recognition analogy; aging erodes this system as cellular mosaicism increases.
Immune 'archetypes' - Krummel's term for recurring configurations of immune-cell behavior that the body runs like a football team's plays; used to explain why autoimmune diseases and cancers come in distinct sub-types requiring different treatments.
Cancer immunotherapy / checkpoint blockade - Raising the threshold at which T-cells activate so they can attack tumors previously being ignored; described as the discovery that reframed immunology from a binary self/foreign system to a tunable one.
Spatial biology / single-cell tissue mapping - Computationally reconstructing how every cell type in a tissue or tumor is 'wired' to every other cell type, used to model what happens if a specific cell-cell signal is disrupted.
Killer experiment - Krummel's term for a well-designed experiment built to disprove a hypothesis, considered more valuable than one designed only to confirm it.
Thymic revitalization / peptide therapy - Emerging area of interest in restoring thymus function (which involutes with age) to boost new T-cell output, discussed alongside unregulated non-FDA-approved thymic peptides people are already self-administering.

Summary

Andrew Huberman hosts Dr. Max Krummel, a UCSF professor of immunology and cancer biology whose thesis work on T-cell checkpoint blockade helped launch modern cancer immunotherapy (and led to his advisor's Nobel Prize), for a wide-ranging tour of how the immune system actually works. Krummel opens with a WWII submarine sonar analogy for self-versus-non-self discrimination, then walks through how immunology's understanding has shifted over his 30-plus year career: from a simple binary system that either ignored the body or attacked foreign invaders, to a highly tunable network that also regulates the microbiome, metabolism, wound healing, and even cardiac muscle turnover.

A major throughline is immune aging, which Krummel frames as having two separate causes. First, ongoing DNA mutation (skin cells alone accumulate 10,000-30,000 mutations per cell per day from UV exposure) turns the body into a genetic mosaic over decades, making it progressively harder for the immune system to tell what's actually dangerous versus merely different, which he ties to rising cancer risk with age. Second, the thymus, the organ that produces and educates new T-cells (discovered to be essential via Jacques Miller's mouse experiments), shrinks dramatically after childhood, cutting the supply of freshly-tuned immune cells. This is the biological basis for current interest in thymic revitalization and unregulated thymic peptides, which Krummel discusses candidly without recommending.

The conversation covers concrete health mechanisms: why sleep deprivation impairs immunity (immune cells retreat to the bone marrow overnight for reparative work; under-eye puffiness after bad sleep is literal lymph fluid pooling), and a striking finding that the brain's insular cortex can encode and later reactivate immune states as part of emotional memory, demonstrated in mice where artificially reactivating neurons tagged during a gut inflammation episode reproduced immune cell accumulation in the gut with no new physical trigger. Krummel connects this to why recalling a stressful or positive memory can genuinely shift measurable immune activity, not just mood.

Roughly the second half of the episode is a long, candid discussion of vaccines, pharma incentives, and trust in science, which Huberman explicitly frames as rare territory for the show. Krummel argues that most vaccine-hesitant parents are not anti-vaccine but are asking legitimate, understudied questions about dosing schedules and combinations, that current protocols are shaped more by convenience and historical single studies than rigorous alternative-schedule testing, and that pharmaceutical companies have little financial incentive to fund such studies (or, separately, to develop biomarker tests that would shrink their own market for checkpoint cancer drugs). He is direct that he vaccinated his own children, delayed one dose based on his read of the underlying immunology, and that the scientific evidence for vaccine harm is weak, while still arguing the timing question deserves a real study rather than dismissal.

The episode closes on cancer treatment strategy and the limits of AI in biological discovery. Krummel argues future disease treatment will increasingly require sequenced, multi-step interventions, nudging biological systems through intermediate states, rather than single-target drugs, because resilient biological systems resist one-lever fixes. He distinguishes machine learning (useful for proposing experiments from existing data) from genuine discovery, which still requires running physical experiments since the answer often isn't present in any existing "corpus." Autoimmune conditions get framed similarly to cancer: as immune "archetypes" run in the wrong context, using psoriasis and the malaria-resistant sickle-cell trait as parallel examples of otherwise-adaptive programs that carry a cost. The episode ends with Krummel discussing why he started writing his Substack, The Immune Beyond, as an effort to make science more relatable and rebuild public trust in scientists as fallible, well-meaning humans rather than a monolithic authority.

Notable Quotes

"It's trying its best. But it's applying the wrong program to the wrong situation." - Dr. Max Krummel

"Nature doesn't necessarily want us banging on it. It'll bang back." - Dr. Max Krummel

"Better living through chemistry still requires better living." - Andrew Huberman

"Any person, scientist, doctor or otherwise, who thinks the way to convince people to change their behavior around vaccines or anything else is to just ram it down the public's throat... is proven to be wrong. The pandemic proved that." - Andrew Huberman